Are Couple-Oriented Interventions Effective Across Chronic Illnesses?: A Reply to Arthurs and Colleagues

We appreciate the opportunity to respond to the commentary of Arthurs and colleagues regarding our review of couple-oriented interventions for chronic illness [1]. Arthurs and colleagues raise four points concerning our meta-analysis. First, the authors note that we did not report effect sizes for each study, as suggested in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [2]. The effect sizes were available at the time of submission, and we are happy to provide this information (see Tables 1, 2, and 3). Second, the authors state that it is not clear how we arrived at our conclusions regarding methodological limitations of the studies that we reviewed and that we did not address risk of bias as suggested in the PRISMA statement. The three limitations that we highlighted were easily discerned from the articles (i.e., lack of intent to treat analyses, lack of information regarding number of intervention sessions attended, and lack of statistical power to detect between-group differences). The proportion of studies with these three limitations ranged from 77% to 84%. Although it can be worthwhile to rate the methodological quality of individual studies along multiple dimensions, this was not a goal of our review. Moreover, the value of incorporating this information in meta-analysis is unclear [3].

Table 1.

Individual effect sizes for meta-analysis of patient depressive symptoms

First author, year	Comparison	Effect size (d)	Standard error	95% CI	p value
Baucom, 2009	COI/usual care	0.35	0.54	−0.72–1.42	0.52
Canada, 2005	COI/POI	0	0.31	−0.59–0.59	1.00
Christensen, 1983	COI/usual care	0.63	0.46	−0.27–1.53	0.17
Fridlund, 1991	COI/usual care	−0.18	0.19	−0.55–0.19	0.34
Giesler, 2005	COI/usual care	0.04	0.22	−0.38–0.47	0.85
Keefe, 2004	COI/POI/usual care	0	0.24	−0.46–0.46	1.00
Keefe, 1996	COI/POI	0.25	0.26	−0.27–0.77	0.34
Kuijer, 2004	COI/usual care	1.11	0.35	0.43–1.79	0.00
Lenz, 2000	COI/usual care	0	0.32	−0.64–0.64	1.00
Manne, 2005	COI/usual care	0.21	0.14	−0.06–0.48	0.12
Martire, 2003	COI/POI	0.56	0.42	−0.26–1.38	0.18
Moore, 1985	COI/POI/usual care	0.15	0.26	−0.36–0.66	0.57
Nezu, 2003	COI/POI	0.23	0.22	−0.20–0.66	0.30
Radojevic, 1992	COI/POI/usual care	0.27	0.26	−0.24–0.78	0.30
Riemsma, 2003	COI/POI/usual care	0	0.13	−0.25–0.25	1.00
Saarijarvi, 1991	COI/usual care	−0.02	0.26	−0.53–0.49	0.94
Scott, 2004	COI/POI/usual care	0.37	0.19	−0.00–0.75	0.05
Thompson, 1990	COI/usual care	0.69	0.27	0.16–1.22	0.01
Turner, 1990	COI/usual care	−0.13	0.33	−0.78–0.52	0.69
van Lankveld, 2004	COI/POI	0.32	0.26	−0.20–0.84	0.23

Please see Martire et al. [1] for description of procedure for computing effect sizes

CI confidence interval, COI couple-oriented psychosocial/behavioral intervention, POI patient-oriented psychosocial/behavioral intervention

Table 2.

Individual effect sizes for meta-analysis of patient marital functioning

First author, year	Comparison	Effect size (d)	Standard error	95% CI	p value
Baucom, 2009	COI/usual care	−0.65	0.55	−1.74–0.44	0.24
Canada, 2005	COI/POI	0	0.30	−0.59–0.59	1.00
Christensen, 1983	COI/usual care	0.34	0.45	−0.54–1.22	0.45
Fridlund, 1991	COI/usual care	0.29	0.20	−0.11–0.69	0.16
Giesler, 2005	COI/usual care	0.14	0.22	−0.29–0.57	0.52
Gortner, 1988	COI/POI	0.03	0.24	−0.45–0.51	0.90
Keefe, 1996	COI/POI	−0.28	0.26	−0.80–0.24	0.29
Keefe, 2004	COI/POI/usual care	0	0.24	−0.46–0.46	1.00
Kuijer, 2004	COI/usual care	0.41	0.32	−0.22–1.04	0.21
Martire, 2003	COI/POI	0.16	0.41	−0.64–0.96	0.70
Martire, 2007	COI/POI	0.40	0.20	0.01–0.79	0.04
Moore, 1985	COI/POI/usual care	0.70	0.30	0.10–1.30	0.02
Northouse, 2007	COI/usual care	0.22	0.13	−0.04–0.48	0.09
Riemsma, 2003	COI/POI/usual care	0.09	0.13	−0.16–0.34	0.48
Saarijarvi, 1991	COI/usual care	−0.03	0.26	−0.55–0.49	0.91
Scott, 2004	COI/POI/usual care	−0.10	0.21	−0.51–0.31	0.63
van Lankveld, 2004	COI/POI	0.42	0.27	−0.10–0.94	0.11
Wing, 1991	COI/POI	0.72	0.32	0.10–1.34	0.02

Please see Martire et al. [1] for description of procedure for computing effect sizes

CI confidence interval, COI couple-oriented psychosocial/behavioral intervention, POI patient-oriented psychosocial/behavioral intervention

Table 3.

Individual effect sizes for meta-analysis of patient pain

First author, year	Comparison	Effect size (d)	Standard error	95% CI	p value
Baucom, 2009	COI/usual care	0.19	0.54	−0.87–1.25	0.73
Fridlund, 1991	COI/usual care	0.17	0.19	−0.20–0.54	0.36
Giesler, 2005	COI/usual care	0.24	0.22	−0.19–0.67	0.27
Keefe, 1996	COI/POI	0.09	0.26	−0.43–0.61	0.73
Keefe, 2004	COI/POI/usual care	0	0.24	−0.46–0.46	1.00
Keefe, 2005	COI/usual care	0.45	0.27	−0.08–0.98	0.10
Martire, 2003	COI/POI	0.39	0.41	−0.42–1.20	0.35
Martire, 2007	COI/POI/usual care	0.31	0.12	0.08–0.54	0.01
Moore, 1985	COI/POI/usual care	0.61	0.31	0.01–1.21	0.05
Radojevic, 1992	COI/POI/usual care	0.18	0.26	−0.33–0.69	0.49
Riemsma, 2003	COI/POI/usual care	0	0.13	−0.24–0.25	0.98
Saarijarvi, 1991	COI/usual care	−0.38	0.26	−0.89–0.13	0.15
Turner, 1990	COI/usual care	0.43	0.33	−0.22–1.08	0.20
van Lankveld, 2004	COI/POI	0.47	0.27	−0.05–0.99	0.08

Please see Martire et al. [1] for description of procedure for computing effect sizes

CI confidence interval, COI couple-oriented psychosocial/behavioral intervention, POI patient-oriented psychosocial/behavioral intervention

The third criticism raised by Arthurs and colleagues is that the broad range of intervention strategies and formats of the studies that we reviewed leaves clinicians with little guidance regarding the best treatment approach and type of patient to target. Indeed, couple-oriented interventions reflect the broader psychosocial intervention literature in terms of diversity in number of components and sessions, as well as type of content (e.g., cognitive-behavioral skills training, education). As we have noted elsewhere, the decomposition of effective interventions for the purpose of identifying effective components is an important goal for future intervention research [4]. However, the couple-oriented intervention literature is not yet at this stage. We would also like to note that our review is not meant to be prescriptive to clinicians. As stated in the original article, our findings are preliminary and in need of corroboration. We do believe that couple-oriented interventions are a promising treatment approach for the many patients who are partnered and that a cross-disease perspective to reviewing this literature and developing new interventions may yield conceptual advances.

Finally, Arthurs and colleagues state that we combined outcomes from different studies without regard to whether those outcomes were specifically targeted by an intervention. The authors provide the example of a weight loss intervention that was not targeted at pain reduction but assessed this outcome and therefore was included in our analysis. In our experience, it is uncommon for researchers to assess change in outcomes that are not targeted in an intervention, and this did not occur in the studies that were included in our review.